Audience research
Part of How do talent agencies carry out audience research in England?
Before you run talent agencies customer interviews, fix the questions
A practical listicle on running talent agencies customer interviews in England: define your audience, ask question-form prompts, and score answers against a rubric.
What to take away
- Most teams start with a script and a wish list, then collect answers that cannot change a decision. Start with the decision instead.
- Interviews are for uncovering needs and language, not for confirming what you already sell. Write prompts as questions, not pitches.
- Recruit a spread of creators, brand contacts and agency staff, then score every transcript against the same rubric so comparisons hold.
- Keep consent, storage and deletion rules visible from the first call, because UK GDPR applies to notes and recordings.
- Use the results to shape roster, pricing and outreach, not to produce a report nobody reads.
Why do most talent agencies customer interviews fail before the first call?
The usual mistake is treating the interview as a sales conversation. The interviewer talks, the participant agrees, and nothing new appears. That feels productive and produces nothing you can act on.
A second failure is recruiting only people who already like you. Existing clients are easy to book and tell you what your current service already does. They cannot tell you why a comparable creator chose another agency.
Fix both by writing the decision first. For example: should we add a short-form editing tier at £600 a month? Then every prompt tests that decision. The wider audience research guide for England sets out how interview work sits alongside desk research and platform data.
Which creators, brands and staff should you interview?
Set inclusion criteria before recruiting. Good criteria cover geography, platform mix, deal size and tenure. If the rule covers England only, say so and record it.
Aim for three groups. Creators on your roster, creators represented elsewhere, and the brand or agency contacts who brief them. Add one internal voice: a booker or contracts lead who sees where deals stall.
Exclude anyone you cannot lawfully process data about, and exclude current negotiations to avoid conflicts. Record why each person was included, so the sample can be explained later.
What should you ask, and what should you avoid?
Ask about past behaviour, not future intentions. "Walk me through the last brand deal you signed" beats "would you pay for this".
Avoid leading prompts, hypothetical pricing and questions that name your own service too early. Save your description for the end, after the participant has described their own process in their words.
When you discuss marketing messages, remember that paid promotion must be recognisable under AdviceOnline guidance for marketers from the ASA and CAP. That shapes what creators can promise a brand.
How should you handle consent, notes and recordings?
Explain what you collect, why, how long you keep it and who sees it. Get consent for recording separately from consent to take part. You can rely on UK GDPR guidance and resources from the ICO when drafting notices and retention rules.
Store recordings in one named folder with access limited to the research lead. Delete at the stated date. Anonymise quotes before they reach a deck.
If a participant raises a complaint about another party, log it without editorialising. Legal context for enforcement sits in the IP enforcement reports published by GOV.UK.
How do you score interviews so the findings hold up?
Score each transcript against fixed criteria. Use a simple 1 to 5 scale and define what each number means before coding starts.
| Criterion | What a low score looks like | What a high score looks like |
|---|---|---|
| Decision relevance | Answer does not touch the decision | Answer changes how you would act |
| Specificity | Opinions and generalities | Named events with dates and amounts |
| Independence | Participant echoes your pitch | Participant describes their own process |
| Corroboration | One-off claim | Matches another interview or a source |
| Actionability | No next step | Names a change to offer or price |
Two reviewers should score the first three transcripts together, then split the rest. Disagreements above one point go to a third reader. Report medians, not averages, when the sample is small.
Keep the scoring sheet next to the transcript. When you compare tools for collecting and storing answers, the survey tools and sources breakdown explains which formats suit small samples.
How do you turn interview findings into agency decisions?
Write one page per decision. State the decision, the evidence, the confidence level and the owner. Two or three strong transcripts can justify a change; one alone cannot.
Give each finding a shelf life. Creator needs move when platform payouts or brand budgets move, so revisit the rubric each quarter rather than treating it as fixed.
Finally, tell participants what changed. That keeps the panel warm for the next round and makes recruitment easier next time.
Common questions
How many interviews are enough for a talent agency?
For a single decision, six to ten well-chosen participants usually surface the main patterns. Beyond that, new transcripts tend to repeat what you already have.
Can we interview creators outside England?
Yes, but keep the sample separate and say which rules apply where. Scotland, Wales and Northern Ireland share some frameworks with England and differ on others, so do not merge them without checking.
Do we need consent to record a call?
Yes. Ask before recording, explain the purpose and retention period, and offer a notes-only alternative for anyone who declines.
Should we pay participants?
There is no single rule. Decide a policy, apply it consistently, and record it in your recruitment notes so the sample can be explained.
